Sent to CCL by: Reinaldo Pis Diez [reinaldo.pisdiez^-^gmail.com]
Dear Peter,
Have a look at a paper by Mark Gordon and others, J Chem Phys 110 (1999)
4199, in which they show how natural orbitals obtained from
single-reference correlated methods can be used a priori to diagnose the
multiconfigurational character of a wavefunction.
Hope this helps.
Regards,
Reinaldo
On 08/26/2011 04:21 AM, Peeter Burk burk::ut.ee wrote:
Sent to CCL by: "Peeter Burk" [burk_._ut.ee]
Dear CAS experts,
I am trying my first CAS calculations and having some problems
connected with active orbital selection (as expected). I understand
that for choosing right orbitals one should have enough knowledge
about the system under study or enough time to try all reasonable (?)
active orbital combinations. I am trying the later approach to get
some experience by trial and error on some simple systems. Perhaps you
can help me with the following questions:
1. Is there a a posteriori indicator that I have included only the
needed orbitals (after I have tried enough combinations)? C. Cramer
has a statement in his Essentials of Computational Chemistry that A
reasonable rule of thumb is that any orbital having an occupation
number greater than 1.98 or less than 0.02 is not important enough to
include in the CAS space. However, I find that even for water (as a
test case at CAS(8,8)/cc-pvdz level) there are 2 virtual orbitals with
occupancies above that threshold (0.246 and 0.236), two occupied
orbitals with occupancies less than 1.98 (1.977 and 1.975), and some
more with occupancies not far from those tresholds. I suspect that for
more complicated systems the picture will be even more unclear...
2. Similarly, is there some a posteriori indicator that the
multideterminant approach was even needed? Chemical intuition (and
wavefunction instability at RHF/UHF level) can be used for predicting
the possible need for CAS treatment, but is there a clear indication
after performing the calculation that CAS was needed (like no
imaginary frequencies is indicator for the minimum on the PES)?
Best regards
Peeter Burk
University of Tartu, Estoniahttp://www.ccl.net/chemistry/sub_unsub.shtmlConferences:
http://server.ccl.net/chemistry/announcements/conferences/>;